AI & Government: Risks of OpenAI, Anthropic Partnerships
- As artificial intelligence rapidly integrates into various sectors, governments are also exploring its potential.
- Anthropic, known for its Claude chatbot, recently unveiled AI models designed for government employees, featuring capabilities to handle classified data and understand bureaucratic language.
- However, experts express concerns about governments becoming overly dependent on these AI models.
AI Use in Government Sparks Ethics debate
As artificial intelligence rapidly integrates into various sectors, governments are also exploring its potential. The U.K. government estimates AI productivity tools could save notable labor time. The U.S. government has documented over 1,700 AI use cases, spurred by federal policies promoting AI adoption and procurement. This surge in government AI use has prompted tech companies to develop specialized models.
Anthropic, known for its Claude chatbot, recently unveiled AI models designed for government employees, featuring capabilities to handle classified data and understand bureaucratic language. These models are already deployed by agencies involved in national security. OpenAI, the creator of ChatGPT, introduced similar government-tailored AI models in January.
However, experts express concerns about governments becoming overly dependent on these AI models. Potential issues include the generation of inaccurate facts, the perpetuation of biases, and the risk of misguided policy decisions. There are also worries about governments becoming locked into specific providers, perhaps leading to increased costs for taxpayers.
“I worry about governments using this kind of technology and relying on tech companies, and in particular, tech companies who have proven to be quite untrustworthy,” said Carissa Véliz, an AI ethics researcher at the University of Oxford.
Véliz highlights the power imbalance, noting that governments are frequently enough playing catch-up in regulating AI technologies. She also raises concerns about financial dependency on companies based in other countries, particularly given geopolitical complexities. This mirrors concerns about over-reliance on platforms like TikTok.
Anthropic says it is committed to transparency, citing published work on model risks and collaborations with governments to test AI systems.
Some observers suggest that the rollout of these models is less about national security and more about reinforcing existing norms within government. Andres Guadamuz,an AI researcher at the University of Sussex,advises governments to proceed cautiously and maintain diverse options in the AI market to secure better negotiating positions.
What’s next
As government agencies continue to explore and implement AI solutions, ongoing evaluation of ethical implications, potential biases, and vendor dependencies will be crucial to ensure responsible and effective use of artificial intelligence in the public sector.